Brintons Portugal is the local factory of UK-based Brintons Limited, one of the largest world manufacturers of high-quality carpets, used in large public spaces such as hotels, casinos, airports and cruise ships. The factory was recently asked to start budgeting new orders, though all commercial and logistics decisions remain centrally made by the group.
A carpet contains three main components: weft, wire and web. Each design is unique per customer specification — colours, patterns, dimensions and intended use. The main factory equipment is the loom, which exists in different versions and can be configured into several alternative manufacturing circuits.
The difficulty with budgeting an order is selecting among the equipment alternatives required to complete it. The best budget implies choosing the production configuration that minimises costs — which requires running different alternatives through the manufacturing process, and at a given time may require considering resources already committed to previous orders.
The Brintons Group developed and patented a new weaving technology — High Definition Weaving (HDW) — to replace the current Electronic Jacquards (EJ), reducing waste and weaving defects. Despite its advantages, HDW is not always financially superior to EJ. The cost factors considered are: energy, operations and maintenance manpower, and total raw materials including waste.
The methodology followed is the Systems Engineering V-Model. The solution follows two main steps: creation of a SysML model representing the costs of the different operations in the production process, and creation of an algorithm that returns the production process cost for a given order. The software used was IBM Rational Rhapsody.
The manufacturing process is simple and sequential, with basically the same steps across alternatives. What makes the costing problem complex is that the choice of equipment at certain steps carries different costs.
The system was modelled using SysML Parametric Diagrams and Sequence Diagrams to describe how requirements, parameters and variables relate to each other. The user inputs are: Order Width, Order Length, Order Quality and Number of Colours. The remaining parameters are fixed and chosen according to the loom type. The intention is to provide cost differences between the 4 looms, rather than asking the user to select one.
From Quality, the system derives Beat-Up, Hair Height, and Yarn Tex and Type. From Width, it determines the number of coils feeding the loom. These feed into the total yarn quantity estimate, which along with energy and labour parameters determines the cost per configuration.
The model was validated using data from 5 real orders: 65674, 68477, 68795, 68809 and 68878. Order 65674, for example, contains five subsets of carpet orders, each with different colours, width, length and quality. The model was deemed successfully validated, and the company is evaluating some of its decisions based on the results obtained.
The model covers approximately 90% of the company's cost reality and 85% of the operations performed to obtain the final product. Some smaller operations with lower or equal costs across alternatives were not included — the focus was on operations with different costs and more significant financial impact. All data used is covered by a non-disclosure agreement.
The dissertation is intended as a specification for the technical team to implement in software. Going forward, the model should be subject to continuous improvement — by the computer team in the architecture, and by the production department in refining the parameters. A Sensitivity Analysis would allow comparing the cost impact of parameter variations, such as evaluating whether investing in auxiliary fault-detection technology on the MK7 EJ loom is financially justified.